Daylight saving time was originally conceived as an energy-saving idea. In 1784 Benjamin Franklin observed that during summer months, people slept during the daylight hours of morning and then burned candles at night for illumination. Thus adjusting schedules to begin earlier in the day during summer months would substitute costly wax for free sunlight.
In later years several governments turned this notion into policy. For example, Germany implemented daylight saving time (DST) during World War I to reduce demand for electric lighting in order to free more coal for the war effort. Since the 1970s DST has become a worldwide phenomenon, with much of the motivation linked to energy conservation. Most recently, the U.S. Energy Policy Act of 2005 extended DST in 2007 by three weeks in the spring and one week in the summer.
All of these policy initiatives are based on the presumption that DST saves energy, but does it?
Two papers published in economics journals suggest DST may actually increase electricity consumption a small amount, or at best serve only to shift electricity use to different parts of the day. For example, Kotchen and Grant (2011) analyzed a unique data set from Indiana – a state with DST time rules that historically have differed across counties. They found that DST increased residential electricity demand by one percent. Similarly, Kellogg and Wolff (2008) analyzed a change in DST rules implemented in Australia as part of preparations for the 2000 Summer Olympic Games. They found that DST does indeed decrease evening electricity use, but it increases use in the morning. Thus no overall energy savings were realized.
These findings are driven by behavioral adjustments. Electric lighting is only one way that people use electricity; other uses depend on how people choose to spend their daylight hours. Kotchen and Grant found for Indiana that DST changed the way that people use air conditioning in the summer, which led to the increase in electricity use. Kellog and Wolff showed that people adjust the timing of their energy use in response to the change in the clock, but do not change overall consumption.